A central issue for hearing instrument wearers is understanding speech in an environment with a plurality of sound sources. A situation of this type arises, by way of example, at a party at which numerous guests are talking at the same time and with approximately the same loudness. This problem is therefore designated as the so-called “cocktail party problem”.
One possibility of improving the intelligibility of speech consists in employing a directional microphone. In this respect, the hearing instrument wearer has to look in the direction from which he wishes to receive sound preferentially. A disadvantage of this solution, however, is the fact that the head must not be turned away from the source. It therefore lacks flexibility. Apart from this, a plurality of sources may lie in the range of angles in which the directional microphone is sensitive.
An alternative method is known from telephone technology. In this case, signals from sources coming from different directions can be separated from the overall signal mix with the aid of so-called BSS technology (blind source separation). By way of example, three speakers are taking part in a telephone conference in one room. It is then possible with BSS technology to make three different channels available on which one speaker is transmitted preferentially in each case.
A method for separating sources is known from the publication US 2003/0185411 A1, in which the signals of different sources can be presented separately. The user can then select the desired source.